Predicting Political Ideology Using Campaign Finance Data

Public interest has become increasingly focused on the role of money in politics. Specifically, the Citizens United v. FEC Supreme Court ruling maintained the legality of unrestricted political expenditures by corporate and union entities. As a result, there has been a proliferation of super PAC, or ‘political action committee’, organizations. While these organizations are not permitted to make contributions directly to campaigns, they may engage in unlimited independent spending and there is no restriction on the amount of funds they can accept from donors. Such legal developments beg the question: Can one predict generally the influence of donors over politicians to whom they give? As the flux of money into the political system continues to increase, this remains a critical question. Our project builds largely on previous research by Poole and Rosenthal and makes use of freely available databases of campaign finance information. We aim to predict quantitative scores of political ideology using campaign contribution profiles; specifically, the input to our algorithm is a matrix of normalized vectors of campaign contribution data, partitioned by industry, the learning algorithm used is Support Vector Regression (SVR), and the output is a score for each candidate in the interval [−1, 1] indicating a position on the political ideological spectrum.